Test.py

import numpy
import matplotlib.pyplot as plt
import json
import cv2
import os
from Detectors import DetectorDay, DetectorNight, DayNightDetector, JapDetector
import Config
from Evaluation import Evaluation
from StableFrameList import StableFrameList
from MaskList import MaskList
from Misc import Image
from AnomalyDetector import AnomalyDetector
import ResultRefinement as rr

#Initilize detector
print('Parse detector result ...')
dayNightDetector = DayNightDetector()
detectorDay = DetectorDay(Config.data_path + '/result_8_3_3_clas.txt', Config.data_path + '/result_8_3_3_nclas.txt')
detectorNight = DetectorNight(Config.data_path + '/extracted-bboxes-dark-videos')
#evalFunc = Evaluation(Config.data_path + '/test_groundtruth.txt')
anomalyDetector = AnomalyDetector()
stableList = StableFrameList(Config.data_path + '/unchanged_scene_periods.json')
maskList = MaskList(Config.data_path + '/masks_refine_v3')

for video_id in range(1, 101):
    print("Processing video ", video_id)
    detector = detectorDay
    if dayNightDetector.checkNight(video_id):
        detector = detectorNight

    stableIntervals = stableList[video_id]
    print(stableIntervals)
    confs = {}
    print(detector.name)

    #anomaly save file
    if not os.path.exists(Config.output_path + '/' + str(video_id)):
        os.makedirs(Config.output_path + '/' + str(video_id))
    f = open(Config.output_path + '/' + str(video_id) + '/anomaly_events.txt', 'w')

    #loop all stable intervals
    for scene_id in range(1, len(stableIntervals) + 1):
        l, r = stableIntervals[scene_id - 1]
        sl = int(l / Config.fps) + 1
        sr = int(r / Config.fps)
        sceneMask = maskList[(video_id, scene_id)]

        #create output folder
        if not os.path.exists(Config.output_path + '/' + str(video_id) + '/' + str(scene_id)):
            os.makedirs(Config.output_path + '/' + str(video_id) + '/' + str(scene_id))

        # output folder: output / video_id / scene_id / stuffs
        # output: average + boxes, gray_boxes before, gray_boxes after mask

        for frame_id in range(sl, sr):
            #print("Frame ID %d" % (frame_id))
            #if frame_id == 15: break
            ave_im = Image.load(Config.data_path + '/average_image/' + str(video_id) + '/average' + str(frame_id) + '.jpg')
            boxes = detector.detect(video_id, frame_id)
            for box in boxes: box.applyMask(sceneMask)

            box_im = Image.addBoxes(ave_im, boxes)

            if detector.name == 'night':
                Image.save(box_im, Config.output_path + '/' + str(video_id) + '/' + str(scene_id) + '/night_average' + format(frame_id, '03d') + '.jpg')

                # save result day model to check if day model is better than night model
                # day_boxes = detectorDay.detect(video_id, frame_id)
                # for box in day_boxes: box.applyMask(sceneMask)
                # day_box_im = Image.addBoxes(ave_im, day_boxes)
                # Image.save(day_box_im, Config.output_path + '/' + str(video_id) + '/' + str(scene_id) + '/day_average' + format(frame_id, '03d') + '.jpg')
            else:
                Image.save(box_im, Config.output_path + '/' + str(video_id) + '/' + str(scene_id) + '/day_average' + format(frame_id, '03d') + '.jpg')

            #detect anomaly event in scene
            anomalyDetector.addBoxes(boxes, frame_id) #input detected boxes => list of anomaly event
            detectedAnomalyEvents, conf = anomalyDetector.examineEvents(video_id, scene_id, frame_id, frame_id == sr - 1, f)

            event_im = anomalyDetector.drawEvents(box_im)

            Image.save(event_im, Config.output_path + '/' + str(video_id) + '/' + str(scene_id) + '/events' + format(frame_id, '03d') + '.jpg')
            confs[frame_id] = conf

    f.close()
    #output anomaly graph text before, anomaly_graph_after, anomaly_graph before, anomaly_graph after, result metric
    print(confs)
    f = plt.figure()
    plt.plot(confs.keys(), [confs[key] for key in confs.keys()])
    # plt.show()
    f.savefig(Config.output_path + '/' + str(video_id) + '/' + str(video_id) + '_anomaly.pdf', bbox_inches='tight')
    plt.close(f)
    f = open(Config.output_path + '/' + str(video_id) + '/' + str(video_id) + '_anomaly.txt', 'w')
    for key in confs.keys():
        f.write(str(key) + ' ' + str(confs[key]) + '\n')
    f.close()

#rr.refineResult(Config.output_path)